Thesis Colloquium Merel Kooi

A 3D groundwater model is developed to better understand rainfall-runoff processes in basaltic catchment. The model is based on field observations from Skaftafellsheiði, Iceland.

Organised by Hydrology and Quantitative Water Management

Tue 21 April 2015 14:15 to 14:45

Venue Forum, gebouwnummer 102
Room C0229

Hydrological processes in a basaltic catchment             

(subtitle) A case study on Iceland

Basaltic aquifers are used to supply drinking water throughout the world. Rainfall-runoff processes in basaltic areas remain poorly studied. Basaltic rock is often characterized by cooling joints, which can transport large amounts of water. Quantifying this water movement is difficult because joints are not easily measured and can be filled up with erosion material over time. A groundwater flow model was developed in order to get more insight in the impact of basalt on rainfall-runoff characteristics. The model was based on field data collected at Skaftafellsheiði, Iceland. Data collection took place in July, August and September 2014. Skaftafellsheiði is a 4 by 6 km basaltic hill, surrounded at the east, south and west by a glacial outwash plain and to the north by the Vatnajökull glacier. Two streams are present in the catchment, on the east and the west side, dividing the catchment in an east and west subcatchment. In the study area, measurements to precipitation, discharge, electric conductivity, water temperature and hydraulic conductivity were done. The field measurements showed a quick response of discharge to precipitation, and a rather constant base flow during dry periods. Discharge could only partly be explained by precipitation, indicating groundwater inflow into the streams. Precipitation was found to increase linearly with surface levels. Calculated potential evapotranspiration was rather high, which can be explained by the 19 potential sun hours per day during summer on Iceland. The amount of water stored in the catchment decreased during the measurement period. A simple 3D MODFLOW GMS model was created for the catchment. MODFLOW was chosen instead of a simple bucket model, because it accounts for 3D spatial variability. Peat, regolith and organic soil formed the upper layer, with a thick basalt layer underneath. The five layer model uses the stream and recharge package. Both a stationary and transient model are made. The sensitivity analysis indicated basalt as the most sensitive parameter of the model. Model calibration was done with respect to discharge measurements at the catchment outlet. The stationary model performed well, using a hydraulic conductivity of 0.3 m d-1 for the basalt aquifer. Deviation of modelled discharge with respect to the average measured discharge at the catchment outlet was 1.5%. Modelled discharges in the headwaters of the catchment were underestimating discharge up to a factor 10 compared to field measurements. The transient model performed moderately well, with a R2 of 0.36 using the same parameter values as for the stationary model. During the measurement period two major rainfall events occurred. The first observed discharge peak was underestimated by the model, while the second was overestimated. Base flow was overestimated too. Correlation between the measured and modelled discharge was 0.67. Spatial variability of wet and dry cells at the surface coincided roughly with field observations for both the stationary and transient model. Peat areas were relatively wet in the field and in the model, though their potential evapotranspiration was high. Peat is present in the flat area of the catchment, were water can accumulate. Modelled groundwater table levels in the upper part of the catchment, which consists of steep slopes and relatively permeable regolith, were meters below the surface. No water accumulation was observed in this part of the catchment. For further studies it is recommended to analyse different scenarios regarding boundary position and type, and geological formations. Hydraulic head and evapotranspiration measurements would increase the hydrological understanding and model quality. Taking into account the simplicity of this model, it can be concluded that rainfall-runoff responses of a basalt catchment can be modelled reasonably accurately in 3D.